mlsanityData to 14 Oct 2026

A Glimpse to Compress: Dynamic Visual Token Pruning for Large Vision-Language Models

arXiv 2508.01548 ↗First posted 3 Aug 2025cs.CVcited by 18

On the map: Multimodal LLMs / vision-language models → Visual token pruning and efficiency

also in 1 more
Abstract (arXiv)

Visual token compression is critical for Large Vision-Language Models (LVLMs) to efficiently process high-resolution inputs. Existing methods that typically adopt fixed compression ratios cannot adapt to scenes of varying complexity, often causing imprecise pruning that discards informative visual tokens and results in degraded model performance. To address this issue, we introduce a dynamic pruning framework, GlimpsePrune, inspired by human cognition. It takes a data-driven ''glimpse'' and prunes irrelevant visual tokens in a single forward pass before answer generation. This approach prunes 92.6% of visual tokens while on average fully retaining the baseline performance on free-form VQA tasks. The reduced computational cost also enables more effective fine-tuning: an enhanced GlimpsePrune+ achieves 110% of the baseline performance while maintaining a similarly high pruning rate. Our work paves a new way for building more powerful and efficient LVLMs.

Key facts. Cited by 18 papers in the map (data through 14 Oct 2026). Placed in Visual token pruning and efficiency, one of 16 directions in Multimodal LLMs / vision-language models. Builds on 46 of its 47 references that are in the map. First posted on arXiv 3 Aug 2025.

Builds on

46 of 47 references.

Closest ancestors · 12

Same direction, newest first

  1. 2025Beyond Attention or Similarity: Maximizing Conditional Diversity for Token Pruning in MLLMs12 Jun
  2. 2025VScan: Rethinking Visual Token Reduction for Efficient Large Vision-Language Models28 May
  3. 2025PACT: Pruning and Clustering-Based Token Reduction for Faster Visual Language Models11 Apr
  4. 2025TopV: Compatible Token Pruning with Inference Time Optimization for Fast and Low-Memory Multimodal Vision Language Model24 Mar
  5. 2025DivPrune: Diversity-based Visual Token Pruning for Large Multimodal Models4 Mar
  6. 2025Dynamic Token Reduction during Generation for Vision Language Models24 Jan
  7. 2024VisionZip: Longer is Better but Not Necessary in Vision Language Models5 Dec
  8. 2024Beyond Text-Visual Attention: Exploiting Visual Cues for Effective Token Pruning in VLMs2 Dec
  9. 2024Dynamic-LLaVA: Efficient Multimodal Large Language Models via Dynamic Vision-language Context Sparsification1 Dec
  10. 2024PyramidDrop: Accelerating Your Large Vision-Language Models via Pyramid Visual Redundancy Reduction22 Oct
  11. 2024Efficient Large Multi-modal Models via Visual Context Compression28 Jun
  12. 2024An Image is Worth 1/2 Tokens After Layer 2: Plug-and-Play Inference Acceleration for Large Vision-Language Models11 Mar

Other references · 34

  1. 2021LoRA: Low-Rank Adaptation of Large Language Modelscited by 14,380
  2. 2024DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Modelscited by 9,379
  3. 2023Visual Instruction Tuningcited by 7,869
  4. 2023BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Modelscited by 6,137
  5. 2025Qwen2.5-VL Technical Reportcited by 5,587
  6. 2024Qwen2.5 Technical Reportcited by 4,942
  7. 2021RoFormer: Enhanced Transformer with Rotary Position Embeddingcited by 4,703
  8. 2023Improved Baselines with Visual Instruction Tuningcited by 4,418
  9. 2024Qwen2-VL: Enhancing Vision-Language Model's Perception of the World at Any Resolutioncited by 4,234
  10. 2022FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awarenesscited by 3,673
  11. 2023FlashAttention-2: Faster Attention with Better Parallelism and Work Partitioningcited by 2,598
  12. 2023InternVL: Scaling up Vision Foundation Models and Aligning for Generic Visual-Linguistic Taskscited by 2,565
22 more
  1. 2016Making the V in VQA Matter: Elevating the Role of Image Understanding in Visual Question Answeringcited by 2,516
  2. 2023Qwen-VL: A Versatile Vision-Language Model for Understanding, Localization, Text Reading, and Beyondcited by 2,224
  3. 2023MMBench: Is Your Multi-modal Model an All-around Player?cited by 2,141
  4. 2022Learn to Explain: Multimodal Reasoning via Thought Chains for Science Question Answeringcited by 1,919
  5. 2025InternVL3: Exploring Advanced Training and Test-Time Recipes for Open-Source Multimodal Modelscited by 1,834
  6. 2023Evaluating Object Hallucination in Large Vision-Language Modelscited by 1,729
  7. 2023MME: A Comprehensive Evaluation Benchmark for Multimodal Large Language Modelscited by 1,652
  8. 2024Expanding Performance Boundaries of Open-Source Multimodal Models with Model, Data, and Test-Time Scalingcited by 1,599
  9. 2024How Far Are We to GPT-4V? Closing the Gap to Commercial Multimodal Models with Open-Source Suitescited by 1,050
  10. 2023SEED-Bench: Benchmarking Multimodal LLMs with Generative Comprehensioncited by 940
  11. 2018VizWiz Grand Challenge: Answering Visual Questions from Blind Peoplecited by 858
  12. 2023LLaMA-VID: An Image is Worth 2 Tokens in Large Language Modelscited by 546
  13. 2023V*: Guided Visual Search as a Core Mechanism in Multimodal LLMscited by 482
  14. 2023Chat-UniVi: Unified Visual Representation Empowers Large Language Models with Image and Video Understandingcited by 371
  15. 2024LMMs-Eval: Reality Check on the Evaluation of Large Multimodal Modelscited by 353
  16. 2024Visual CoT: Advancing Multi-Modal Language Models with a Comprehensive Dataset and Benchmark for Chain-of-Thought Reasoningcited by 280
  17. 2024Enhancing the Reasoning Ability of Multimodal Large Language Models via Mixed Preference Optimizationcited by 239
  18. 2025MLLMs Know Where to Look: Training-free Perception of Small Visual Details with Multimodal LLMscited by 176
  19. 2024DyCoke: Dynamic Compression of Tokens for Fast Video Large Language Modelscited by 96
  20. 2024ReTaKe: Reducing Temporal and Knowledge Redundancy for Long Video Understandingcited by 33
  21. 2025LLaVA-Scissor: Token Compression with Semantic Connected Components for Video LLMscited by 26
  22. 2024Free Video-LLM: Prompt-guided Visual Perception for Efficient Training-free Video LLMscited by 17

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 2; 2026: 16.

18 papers.

Sort:

By area:

  1. 2026The Model Knows Which Tokens Matter: Automatic Token Selection via Noise Gatingcited by 11
  2. 2026Beyond Surrogate Gradients: Fully Differentiable Token Pruning for Vision-Language Modelscited by 7
  3. 2026Improving Visual Reasoning with Iterative Evidence Refinementcited by 5
  4. 2026Stepwise Token Selection for Efficient Multimodal Large Language Modelscited by 5
  5. 2026Accelerating Multimodal Large Language Models with Prior-Corrected Token Reductioncited by 3
  6. 2025ZSPAPrune: Zero-Shot Prompt-Aware Token Pruning for Vision-Language Modelscited by 2
  7. 2026See What I Mean: Aligning Vision and Language Representations for Video Fine-grained Object Understandingcited by 2
  8. 2026WaveZip: Wavelet-Driven Space-Time Decoupling for Video Token Condensationcited by 2
  9. 2026RUTA: Principled Visual Token Allocation via Rate-Utility Optimizationcited by 2
  10. 2026Semantic-Geometric Dual Compression: Training-Free Visual Token Reduction for Ultra-High-Resolution Remote Sensing Understandingcited by 1
  11. 2026Spectral Evolution-Guided Token Pruning in Multimodal Large Language Modelscited by 1
  12. 2026DiffPrune: differentiable information throttling for token pruning in vision-language modelscited by 1
6 more
  1. 2025ShaRP: SHAllow-LayeR Pruning for Efficient Video Large Language Modelscited by 0
  2. 2026Predictive Regularization Against Visual Representation Degradation in Multimodal Large Language Modelscited by 0
  3. 2026VLZip: Unified Visual and Textual Compression for Interleaved Long-Context Modelingcited by 0
  4. 2026StepPrune: Adaptive Sequential Visual Token Selection across Multimodal Large Language Modelscited by 0
  5. 2026SCOPD: Sparse-Context On-Policy Self-Distillation for Efficient Vision-Language Modelscited by 0
  6. 2026V-CoLA: Vision Token Compression with Linear Attentioncited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025GreedyPrune: Retenting Critical Visual Token Set for Large Vision Language Models
  2. 2026SlimVLM: Sensitivity-aware Dynamic Structured Pruning with Adaptive Visual Token Selection for Efficient Vision-Language Models
  3. 2025Balanced Token Pruning: Accelerating Vision Language Models Beyond Local Optimization
  4. 2026AsymVLM: Asymmetric Token Pruning for Efficient Vision-Language Model Inference
  5. 2026Prune Redundancy, Preserve Essence: Vision Token Compression in VLMs via Synergistic Importance-Diversity
  6. 2025ToDRE: Effective Visual Token Pruning via Token Diversity and Task Relevance
  7. 2025CoViPAL: Layer-wise Contextualized Visual Token Pruning for Large Vision-Language Models
  8. 2026LearnPruner: Rethinking Attention-based Token Pruning in Vision Language Models